A national study of the prevalence of HIV antibody designed to monitor sexual spread of HIV infection in England and Wales was made of homosexual and heterosexual patients attending sexually transmitted disease clinics in four districts in 1985, seven in 1986, and 14 in 1987. Patients were invited to participate and were counselled. Among homosexual men in two clinics in south east England, HIV antibody was found in 92 (12.9%) of 711 in 1985, 65 (15.2%) of 428 in 1986, and 81 (14.6%) of 556 in 1987: corresponding findings in the other regions were 16 (5.0%) of 321, 41 (6.3%) of 654, and 21 (3.1%) of 678. The prevalence of HIV antibody was higher in homosexual than bisexual men, in patients aged 25 years or more, or with one or more specified minor complaints. Among heterosexual patients in the south east in 1986, HIV antibody was found in seven (3.0%) of 230 men and three (1.3%) of 233 women and in 1987 in 10 (1.0%) of 962 men and seven (0.7%) of 949 women. In other areas corresponding findings in 1986 were two (0.2%) of 950 men and three (0.4%) of 752 women and in 1987 were three (0.06%) of 5312 men and one (0.02%) of 4778 women. All but one of the heterosexual patients with the antibody were intravenous drug abusers or had had sexual contacts in or were from an area abroad with a high prevalence of AIDS. Failure to identify a heterosexual patient with HIV antibody not in a risk group (other than that of being an attender at the clinic) or who did not have a sexual partner in a risk group suggests that their prevalence in the patient population of the clinics in the south east is less than one in 700 and in the other regions less than one in 3000. Refusals to participate increased during the study but comparisons of patients who agreed and refused in terms of age, the presence of symptoms suggesting AIDS, travel abroad, and number of sexual partners a month showed little evidence of selective bias.